86_ftp team mental models in a team knowledge

Journal of Organizational Behavior

J. Organiz. Behav. 22, 89106 (2001)

Team mental models in a team knowledgeframework: expanding theoryand measurement acrossdisciplinary boundaries

SUSAN MOHAMMED* AND BRAD C. DUMVILLEThe Pennsylvania State University, Department of Psychology, Pennsylvania, U.S.A.

Summary Because research on team mental models is still in its formative stages, there is aneed for continued conceptual development of the construct and direct empiricalsupport linking team mental models to team outcomes. Researchers in other eldshave developed concepts that are distinct from, but clearly related to team mentalmodels, including information sharing, transactive memory, group learning, andcognitive consensus. Although these research streams currently exist in parallel withlittle cross-fertilization, there is much to be gained from integration across disciplin-ary boundaries. Therefore, the purpose of this paper is to enrich the theoreticalunderstanding of team mental models and to broaden the empirical research baseby adopting a cross-disciplinary focus and incorporating related team knowledgedomains from other literatures. Based on a synthesis of various literatures, wedevelop a framework that delineates the relationships among team knowledge con-structs. Copyright # 2001 John Wiley & Sons, Ltd.

Introduction

The notion of a team mental model was developed to help account for performance differences

between teams (Cannon-Bowers and Salas, 1990 Paper presented at the SIOP, Miami, FL; Rouse

et al., 1992) and refers to an organized understanding of relevant knowledge that is shared by team

members (Cannon-Bowers et al., 1993; Klimoski and Mohammed, 1994). The general thesis of the

shared mental model literature is that team effectiveness will improve if team members have an ade-

quate shared understanding of the task, team, equipment, and situation (e.g., Duncan et al., 1996). As

this research is still in its formative stages, there is a need for continued conceptual development of the

construct and direct empirical support linking team mental models to team outcomes.

Toward this end, the purpose of the current paper is to enrich the theoretical understanding of team

mental models and to broaden the empirical research base by adopting a cross-disciplinary focus and

incorporating related team knowledge concepts from other literatures. Although team mental models

Copyright # 2001 John Wiley & Sons, Ltd.

* Correspondence to: Susan Mohammed and Brad C. Dumville The Pennsylvania State University, Department of Psychology,415 Bruce V. Moore Building, University Park, PA 16802-3104 U.S.A.

have been investigated primarily by Industrial/Organizational psychologists, researchers in other

elds, including social psychology, cognitive psychology, decision making, and organizational beha-

vior have developed concepts that are distinct from, but clearly related to, team mental models. These

include information sharing, transactive memory, group learning, and cognitive consensus.1 Despite

the relevance of these topics to mental models, they are not typically cited in the team mental model

literature, nor do these literatures commonly cite the existing work on team mental models. Neverthe-

less, our understanding of team mental models can be signicantly enhanced by recognizing, review-

ing, and integrating these separate research streams on team knowledge from other disciplines.

The paper is organized in the following manner. After briey reviewing the literature on team men-

tal models, we summarize the theoretical and empirical research on information sharing, transactive

memory, group learning, and cognitive consensus. Following each summary, we discuss the implica-

tions of the research for enhancing our understanding of team mental models and suggest potentially

fruitful research directions. We then present an integrative synthesis of these parallel research streams.

Team Mental Model Summary

The notion of a shared mental model was introduced by Cannon-Bowers and Salas (1990 Paper pre-

sented at the SIOP, meeting, Miami, FL) to account for the uid, implicit coordination frequently

observed in effective teams and to advance the understanding of how teams function in complex,

dynamic, and ambiguous situations. Team mental models are team members' shared, organized under-

standing and mental representation of knowledge about key elements of the team's relevant environ-

ment (Klimoski and Mohammed, 1994).

Theoretical development

Team mental model content includes shared representations of tasks, equipment, working relation-

ships, and situations (Cannon-Bowers et al., 1993; Duncan et al., 1996; Rouse et al., 1992). In addi-

tion, various types of knowledge are represented, such as declarative (knowledge of what), procedural

(knowledge of how), and strategic (knowledge of the context and application) (Converse and Kahler,

1992 unpublished manuscript). Team mental models also fulll multiple purposes, including descrip-

tion, prediction, and explanation (Rouse et al., 1992). Apart from their essential nature in terms of

content, form of representation, and purpose, Kraiger and Wenzel (1997) suggest four categories of

determinants: environmental, organizational, team, and individual. With regard to consequences, team

mental models bring explanatory power to team performance by directly impacting team processes and

enabling members to formulate accurate teamwork and taskwork predictions (e.g., Cannon-Bowers

et al., 1993; Klimoski and Mohammed, 1994).

1Adopting a multidisciplinary approach, we selected information sharing, transactive memory, group learning, and cognitiveconsensus because they were potentially useful in the study of team mental models, although not commonly referenced in thisliterature. Although space constraints forced us to be selective, we acknowledge that there are several other literatures that areclearly relevant to team mental models (e.g., team situation awareness, team decesion making, language psychology, computer-mediated communication). In addition, we recognize that many of the team knowledge constructs reviewed in this paper havecounterparts at the organizational-level of analysis. For example, literatures exist on organizational memory (e.g., Walsh andUngson, 1991), organizational learning (e.g., Miller, 1996), and organizational frames / issue interpretation (e.g., Daft and Weick,1984). However, this paper focuses on group-level research as opposed to organizational-level research.

90 S. MOHAMMED AND B. C. DUMVILLE

Copyright # 2001 John Wiley & Sons, Ltd. J. Organiz. Behav. 22, 89106 (2001)

Empirical research

Although there have been several theoretical papers describing team mental models (e.g., Cannon-

Bowers et al., 1993; Klimoski and Mohammed, 1994; Kraiger and Wenzel, 1997; Rentsch and Hall,

1994), empirical work has substantially lagged behind conceptual development. Two reasons offered

for this paucity of empirical work include a lack of adequate conceptual development of the construct

and confusion over how to measure cognitive structures at the group level (Mohammed et al., 2000).

Thus far, the most common methodologies for determining mental model content are similarity ratings

(e.g., Mathieu et al., 2000; Stout et al., 1999) and Likert-scale questionnaires (e.g., Blickensderfer

et al., 2000 Paper presented at the APS, Miami, FL). On the other hand, the techniques used most

often to capture the relationships between elements in a person's mind include Pathnder (e.g., Stout

et al., 1999), UCINET (e.g., Mathieu et al., 2000), and concept mapping (e.g., Marks, 1997). As there

is no one best technique, researchers must justify their choice of measurement after carefully consid-

ering the research question and team context (Mohammed et al., 2000). Because of the complexity and

multidimensional nature of team mental models, researchers propose that multiple measures are neces-

sary for thorough assessment (e.g., Kraiger and Wenzel, 1997).

Although much of the earlier research simply used team mental models as a post hoc explanation for

performance differences between teams (e.g., Kleinman and Serfaty, 1989), recent work is attempting

to measure the construct more directly. For example, Mathieu et al. (2000) examined the effect of

shared mental model convergence on team processes and performance using two-person, undergraduate

teams performing a PC-based ight/combat simulation. Results indicated that both teamwork and task-

work mental models related positively to team process and performance, and that team processes fully

mediated the relationship between shared mental models and performance (Mathieu et al., 2000). Other

research has found that team planning (Stout et al., 1999), self-correction training (Blickensderfer et al.,

1997), and computer-based instruction (Smith-Jentsch et al., 1999 Paper presented at the 14th SIOP

conference, Atlanta, GA) foster the development of team mental model convergence.

Because the existing work on team mental models is in its infancy, the current paper seeks to enrich

and expand theoretical and empirical research in this area by incorporating team knowledge concepts

from other disciplines. Toward this end, the literatures on information sharing, transactive memory,

group learning, and cognitive consensus will be reviewed.

Information Sharing Summary

Introduced by Stasser and Titus (1985), the information sharing literature examines information pool-

ing behaviors in groups. In principle, groups are capable of producing better decisions by pooling infor-

mation. In practice, however, collective sampling dynamics promote the rehashing of shared information

at the expense of pooling unshared information (e.g., Gigone and Hastie, 1993; Stasser et al., 1989).


Research in the area of information sharing revolves around the biased sampling model of group dis-

cussion, which is restricted to unstructured, face-to-face discussion with the goal of consensus (Stasser

and Titus, 1985, 1987). According to the model, discussion is biased in favor of shared information, as

shared information is more likely to enter discussion than unshared information. Shared information is

held by all group members before discussion begins, whereas unshared information is only held by one

TEAM MENTAL MODELS IN A TEAM KNOWLEDGE FRAMEWORK 91


group member prior to discussion (Stasser and Titus, 1985). Thus, in this literature, the terms shared

and unshared refer only to the distribution of the information prior to discussion, and do not imply that

information designated as such will or will not be shared during discussion. Taking its limitations into

account, the information sampling model has been shown to be successful when equal-status members

have comparable information loads, shared and unshared items are similar in content and task rele-

vance, and no cues in the decision environment allow members to distinguish shared from unshared

items (e.g., Stasser and Titus, 1985, 1987; Stasser et al., 1989).

Empirical research

Following the prototype of Stasser and Titus' (1985, 1987) seminal work, information sharing effects

have been primarily examined using a hypothetical, hidden prole task (i. e., murder mystery; select

the best candidate), where information that supports a superior option is unshared, whereas informa-

tion that supports an inferior option is predominantly shared. The hidden prole and the best decision

have the potential to be discovered, but only if group members pool their unshared information effec-

tively. Suggesting a robust effect, numerous studies have supported the bias of shared information over

unshared information (e.g., Gigone and Hastie, 1993; Stasser et al., 1989; Stasser et al., 1995). Shared

information not only has a sampling advantage, but it also has a repetition and recall advantage over

unshared information (Larson et al., 1996; Stasser et al., 1989; Stewart and Stasser, 1998).

Temporal, social, and task environment variables are three categories of effects that have been found

to inuence information sharing (Wittenbaum and Stasser, 1996). With regard to temporal patterns,

shared information tends to be discussed earlier than unshared information and information dis-

cussed later in the discussion frequently has less impact on the decision (e.g., Larson et al., 1994).

Social variables affecting information sharing include leadership (e.g., Larson et al., 1996, 1998),

status (Hollingshead, 1996a), and expertise (Stasser et al., 1995). In addition, various aspects of the

task environment have been found to impact the information sharing process, including information

load (e.g., Stasser and Titus, 1987), intellective versus judgement tasks (e.g., Stasser and Stewart,

1992), and recall versus choice demands (e.g., Hollingshead, 1996b; Stewart and Stasser, 1998).

Overview

Over the past 15 years, the information sharing literature has produced a rather homogeneous body of

empirical research showing that face-to-face groups are better at discussing widely shared information

than they are at pooling diverse information. Although the collective information sampling model

(Stasser and Titus, 1985, 1987) provided a solid foundation for theoretical development in this litera-

ture, there have been few conceptual advances beyond the original model. Therefore, theoretical work

has lagged behind empirical ndings, and an advanced framework would be helpful in describing the

interpersonal dynamics of group discussion and interconnecting the nearly two dozen variables inu-

encing information sharing. As most studies are direct variations of Stasser and Titus (1985, 1987) and

use intellective, hidden prole tasks in ad hoc laboratory groups, there is a need to expand the research

to intact organizational teams. In addition, researchers should examine how frequently hidden proles

occur in eld settings.

Information sharing and team mental models

A number of important differences exist between the literatures on information sharing and team

mental models. Because information sampling focuses exclusively on the discussion of information,



the primary dependent variables of interest are the number of times shared/unshared information is

mentioned/repeated, free recall, and the decision that is reached. In contrast, team mental model

research broadens beyond decision making to examine the impact of knowledge convergence on var-

ious team processes and team performance. In addition, most of the information sharing research uses

`extreme' distributions in which either all information is shared by all group members or unshared

information is held by only one member. Rather than being dichotomous in orientation, the team men-

tal model literature focuses on the degrees of sharing that often exist between completely identical and

completely idiosyncratic cognitive representations. Moreover, whereas information sharing addresses

the communication of distinct pieces of information, the emphasis of mental models is on the organi-

zation and interconnectedness of items of information. That is, measures of team mental models not

only reveal the degree of convergence among members with regard to the content of known elements,

but the structure of data or the relationships between elements (e.g., Mohammed et al., 2000).

Despite these differences, there are several ways in which the information sharing research can con-

tribute to an enriched understanding of team mental models. Although the theoretical work on team

mental models does acknowledge the potential benets of diverse structures (e.g., Cannon-Bowers

et al., 1993), the empirical emphasis has been on the convergence or similarity of team member know-

ledge (e.g., Mathieu et al., 2000). However, the work on information sharing warns of the danger of

focusing exclusively on shared information. Distinct team roles often require that members possess

unique information, and this information should not only be mentioned, but also actively considered.

In order to experience the benets of group membership, individuals must have mutually recognized

and complimentary domains of expertise. Therefore, team mental model research should give

increased attention to what is known about effective information pooling behaviors and the factors that

undermine the opportunity for group members to discuss diverse knowledge.

Team mental model studies could also benet from greater emphasis on group discussion and the

dynamics of information exchange. Utilizing various analysis techniques, aircrew communications

research has found that communication sequences differ between effective and ineffective ight crews

(e.g., Bowers et al., 1998). Thus, information sharing and team communications research could be

combined to better explain how team members develop accurate expectations of team behavior. In

addition, whereas much of the information sharing empirical research examines the variables that

increase the probability that unshared information will be mentioned, existing team mental model

work has not focused on antecedents or developmental processes. However, potentially fruitful direc-

tions include examining the impact of temporal environment, leadership, member status, and task

environment on mental model convergence. Furthermore, as team mental model research is in its

infancy, it could benet from the example of systematic, programmatic research utilizing a common

task and a similar research design. Although failure to expand to new samples, tasks, and research

approaches can lead to an empirical rut over time, homogeneity in methodology and a well-accepted

paradigm early on can be benecial in launching an area of research.

Transactive Memory Summary

Transactive memory is a relatively new concept that was introduced by Wegner (1987) to explain

aspects of the behavior of intimate couples (e.g., Wegner et al., 1991), but has also been recently

applied to groups (e.g., Moreland, 2000). It refers to the idea that memory is a social phenomenon,

and individuals in continuing relationships often utilize each other as external memory aids to supple-

ment their own limited and unreliable memories.




Transactive memory is a set of individual memory systems which combines the knowledge possessed

by particular members with a shared awareness of who knows what (Wegner, 1987). The analogy is

that of a computer network in which each computer memory contains its own directory and also a

directory of the other computer memories in the system (Wegner, 1995). A cognitively interdependent

system for encoding, storing, and retrieving information is developed to ensure that important details

are not forgotten (Wegner, 1987). That is, each member keeps current on who knows what, channels

incoming information to the appropriate person, and has a strategy for accessing the information. In

addition to knowing who is the expert in different knowledge areas, transactive memory also involves

storing incoming information with individuals who have matching domains of expertise and accessing

relevant material from others in the system (Wegner, 1987, 1995). Through the encoding and informa-

tion allocation processes, individual memories become progressively more specialized and are fash-

ioned into a differentiated collective memory that is useful to the group.

The specialization of knowledge that individuals develop within a transactive memory system can

reduce the cognitive load of each individual, provide access to an expanded pool of expertise, and

decrease redundancy of effort (Hollingshead, 1998b). On the other hand, the complexity of transactive

memory can create confusion, especially when expertise is in dispute and important information falls

through the cracks (Wegner, 1987).

Empirical research

Because the notion of a transactive memory system was introduced to explain aspects of the behavior

of intimate couples, much of the empirical research has continued in the tradition of examining dyads

(e.g., Hollingshead, 1998a, 1998b; Wegner et al., 1991). However, Moreland and colleagues have con-

ducted a series of experiments investigating the operation of transactive memory in laboratory work

groups. In the rst experiment (Liang et al., 1995), undergraduates were trained to assemble a radio

either individually or in groups and were later tested with their original group or in a newly formed

group. Evidencing stronger transactive memory systems, members of groups trained together specia-

lized in remembering different aspects of the task, coordinated behaviors more effectively, and dis-

played greater trust in each other's expertise. In addition, the effects of group training on task

performance were mediated by the operation of transactive memory. In follow-up work utilizing a

similar design and task, transactive memory was measured more directly through the complexity of

group members' beliefs about one another's radio expertise, the accuracy of those beliefs, and within-

group agreement about the distribution of expertise (Moreland, 2000). Addressing the need to inves-

tigate the developmental process of transactive memory, qualitative research has also examined the

sequence by which encoding unfolds over time (Rulke and Rau, 1997). Recent work has also begun

to validate eld measures and establish the link between transactive memory and team performance in

organizational settings (e.g., Lewis, 2000 Paper presented at the Academy of Management, Toronto).

Overview

In contrast to the information sharing literature, empirical research on transactive memory lags behind

theoretical development. The current work has provided evidence of the existence of transactive mem-

ory (Moreland, 2000; Wegner et al., 1991) as well as the encoding, storage, and retrieval mechanisms

through which it operates (Hollingshead, 1998a, 1998b; Rulke and Rau, 1997). However, much of this



research involves couples rather than groups and has been conducted in the laboratory using contrived

tasks. Recent work has begun to utilize organizational teams, but more research examining how trans-

active memory systems emerge and are maintained in eld contexts is needed.

Transactive memory, information sharing, and team mental models

Although there is not much cross-fertilization across the three literatures, the notion of transactive

memory is clearly related to the construct of a team mental model and information sharing behaviors

in groups. In order to overcome some of the barriers to communicating and using unique knowledge,

Stasser (1991) recommends that each group member be made aware that he or she may have distinctive

information and that members be informed of the types of unique information that others may have.

Therefore, the development of a transactive memory system may overcome the tendency in groups to

rehash shared information at the expense of pooling unshared information.

Whereas transactive memory emphasizes task-oriented domains of expertise, team mental models

incorporate a wider array of content, including shared representations of working relationships and

teamwork. In addition, because of the focus on the encoding, storage, and retrieval of information,

recall is usually the primary dependent variable measured in empirical work on transactive memory.

In contrast, the emphasis in team mental model research is on examining the impact of knowledge

convergence on team processes (e.g., communication, coordination, performance monitoring) and per-

formance.

To date, team mental model research has emphasized the convergence/similarity/agreement/com-

patibility/overlap among team member knowledge as an important predictor of team effectiveness

(e.g., Cannon-Bowers et al., 1993; Mathieu et al., 2000). However, the denition of sharing encom-

passes not only having in common (e.g., share the equipment), but dividing (e.g., share the workload)

(Klimoski and Mohammed, 1994). That is, shared knowledge incorporates both overlapping and com-

plimentary perspectives. According to Cooke et al. (2000), the latter denition of sharing is most

appropriate for heterogeneous teams where distinct team roles require information unique to particular

individuals. Clearly, many teams fall into this category. On a legal defense team, for example, there is

information and areas of expertise that are held in common by all team members, but there is also

information and expertise that is distributed unevenly within the team.

Similar to the notion of transactive memory, some team mental model researchers have discussed

the importance of knowing who knows what, given the distribution of roles within a team. For exam-

ple, Cooke et al. (2000) emphasize the need for measures of interpositional accuracy (accuracy asso-

ciated with roles other than your own) and knowledge distribution metrics (gaps in individual

knowledge can be compensated for by the knowledge of other team members). In addition, Jenkins

and Rentsch (1995 Symposium presented at the SIOP meeting, Orlando, FL) have examined schema

accuracy (accurate knowledge about another's knowledge). However, the team mental model litera-

ture, by and large, has overemphasized the overlapping perspective of sharing and underemphasized

the complementary perspective of sharing. The majority of the empirical research examines the simi-

larity among all team members' understanding of the relationships between concepts. Interestingly,

several of these studies have failed to nd signicant relationships between measures of convergence

of mental models and various dimensions of team performance (e.g., Adelman et al., 1986; Blickens-

derfer et al., 2000 Paper presented at the APS, Miami, FL; Stout et al., 1999).

Perhaps one reason for these weak linkages is the failure to consider the distributed denition of

sharing. Overlapping knowledge in teams with distinct roles may be inefcient, create a redundancy

of effort, and contribute to a less than optimal use of resources. Therefore, rather than measuring simi-

larity globally and assuming that all team members need to have common knowledge in all domains,



future research should work toward specifying the domains and conditions under which distributed and

common knowledge will aid or hinder team performance. In addition, in situations where overlapping

knowledge is required, it is important to delineate which team roles need to share what information. In

a surgical team, for example, there will be some knowledge that needs to be held in common by all

team members (identical), some knowledge that needs to overlap among various dyads or triads

(e.g., nurse and surgeon, surgeon and anaesthesiologist), and some knowledge that will be unique

to individual roles within the team (complementary).

In addition to an increased emphasis on the distributed denition of sharing, team mental model

research could also benet from transactive memory's attention to the basic cognitive mechanisms

of the encoding, storage, and retrieval of information. Although it has been identied as an important

dependent variable (e.g., Mathieu et al., 2000), accuracy is measured to a greater extent in the empiri-

cal research on transactive memory as compared to team mental models. Furthermore, it may

prove useful to investigate turnover (Moreland, 2000) and communication media (e.g., face-to-face

versus computer networked environments, Hollingshead, 1998a) as antecedents of team mental model

development.

Group Learning Summary

In recent years, the concept of organizational learning (e.g., Miller, 1996) has grown in popularity. In

addition, a related, but distinct literature has emerged on group or collective learning, which involves

the construction of new knowledge by a group. In contrast to research at the organizational level, stu-

dies of group learning focus on internal group processes such as communication and inuence in

mostly laboratory settings (e.g., Argote et al., 1999).


Group learning is dened in terms of both processes and outcomes of group interaction (Argote et al.,

1999). Emphasizing processes, Edmondson (1999) conceptualizes learning as `an `ongoing process of

reection and action, characterized by asking questions, seeking feedback, experimenting, reecting

on results, and discussing errors' (p. 353). As an outcome, group learning refers to relatively perma-

nent changes in the knowledge and performance of an interdependent set of individuals associated with

experience (Devadas and Argote, 1995 Paper presented at the midwestern Psychological Associa-

tion, Chicago). For example, Argote et al. (1995) examined learning curves, which refer to groups and

organizations improving productivity as they gain experience in production. Because not all knowl-

edge manifests itself in performance changes, and knowledge is difcult to measure directly, both

changes in knowledge and behavior are useful indicators of group learning (Devadas and Argote,

1995 Midwestern Psychological Association). Highlighting the paradoxical nature of the group

learning process, a group of skilled individual learners will not necessarily result in a learning team

in which collective learning occurs (Argote et al., 1999).

Edmondson (1999) proposed a model of team learning in which psychological safety (shared belief

that the team is safe for interpersonal risk taking) contributes to team learning behaviors, such as seek-

ing feedback, sharing information, experimenting, asking for help, and talking about errors. These

learning behaviors, in turn, facilitate effective performance by allowing the team to shift directions

as situations change and discover unexpected implications of team actions (Edmondson, 1999).



Adopting a different approach, Argote et al. (1999) embed the notion of collective learning within a

CORE framework describing Construction, Operation, Reconstruction, and External relations pro-

cesses in groups. Viewed from this perspective, learning plays a key role across a group's life span, from

initial establishment to task execution to project modication to the management of external sources.

Empirical research

There are only a few examples of empirical research specically investigating group learning. Argote

et al. (1995) examined the effects of turnover and task complexity on group learning in a laboratory

setting. Supporting the existence of a group-level learning curve, the performance of the groups mak-

ing origami birds increased signicantly over the six periods, and this increase in performance

occurred at a decreasing rate. In addition, turnover and task complexity were detrimental to perfor-

mance, and the differences between turnover and no turnover groups as well as simple task and com-

plex task groups were amplied as the groups gained experience over time. In a follow-up study using

a similar experimental paradigm, Devadas and Argote (1995 Midwestern Psychological Association)

replicated previous results regarding a group-level learning curve and turnover, but also examined task

structure. When there was turnover, low structure groups performed less well than high structure

groups because the low structure groups had more difculty accessing existing knowledge, lost knowl-

edge, and had to keep reorganizing when members left the group. Using organizational work teams,

Edmondson (1999) tested her model of team learning using both quantitative (surveys) and qualitative

(interviews and observations) methods of data collection. As hypothesized, team psychological safety

positively affected learning behaviors, which in turn, positively affected team performance. A follow-

up eld study also found that learning oriented beliefs promoted group performance and that effective

coaching, clear direction, and a supportive work context were antecedents of group learning (Cannon

and Edmondson, 2000 Paper presented at the Academy of Management Conference, Toronto,

Canada).

Overview

Although primarily laboratory based, recent work on group learning is beginning to explore eld and

qualitative applications (e.g., Edmondson, 1999). However, as this literature is in its formative stages,

there are a number of areas in need of increased attention. Given the multidimensional and complex

nature of the construct, further theoretical development is clearly warranted as to what constitutes

group learning. Although claiming to measure the same phenomenon, the lack of overlap between con-

ceptual models is quite striking (e.g., Argote et al., 1999; Edmondson, 1999). In addition, the distinc-

tions made between antecedent variables, contextual factors, and components of the construct itself are

ambiguous. In contrast to the homogeneity of the information sharing literature, the group learning

literature suffers from the problem of insufcient cohesion. Greater consensus in the development

of a theoretical framework would be helpful in generating more empirical research.

Group Learning and Team Mental Models

Failure to share unique information and develop a transactive memory system creates problems for

learning in groups. In addition, in order for a team to achieve a shared, organized understanding of



knowledge about key elements in the relevant environment, changes in the knowledge and/or behavior

of team members will most likely occur. Therefore, group learning plays a signicant role in the devel-

opment, modication, and reinforcement of mental models and can be viewed as a sub-set of the

broader concept of team mental models.

Similar to Edmondson's (1999) consideration of corrective feedback and group reection as critical

team learning behaviors, team self-correction has been identied as an important contributor of team

mental models (e.g., Blickensderfer et al., 1997). Specically, team self-correction refers to the natural

tendency of a team to debrief following a performance event by correcting team attitudes, behaviors,

and cognitions (Blickensderfer et al., 1997). Clearly, the error identication, problem solving and feed-

back that characterize team self-correction serve as important resources for team learning. However,

team mental model research could benet from increased attention to the important role that group

learning plays in the development of a team mental model.

Whereas the information sharing and transactive memory literatures emphasize the distributed de-

nition of sharing, the group learning and team mental model literatures emphasize the `held in com-

mon' denition of sharing. The issue of whether the learning of a few, the majority, or all team

members constitutes group learning is usually not discussed explicitly, and team mental model

research has wrestled with similar issues. Consequently, there is still progress to be made in specifying

exactly what constitutes a shared understanding of team-related phenomenon. Requiring all team

members to learn all aspects of mental model content would undermine several of the benets of con-

structing a team in the rst place, including a division of labor and diversity of expertise. Rather, in

order to achieve a high level of team effectiveness, there is some knowledge that will need to be col-

lectively learned by all team members, some knowledge that will need to overlap among various dyads

or triads, and some knowledge that will uniquely be held by particular individuals within the team.

Therefore, future research must begin to specify which team knowledge content needs to be identical,

overlapping, and distributed among team members for maximum team performance. As the optimum

mix of complementary and overlapping knowledge will depend on several factors, researchers should

carefully consider variables such as the team context and team task.

Cognitive Consensus Summary

Many groups such as governing boards, cross-functional teams, and task forces involve the participa-

tion of decision makers from diverse functional backgrounds, multiple departments, and organiza-

tional levels. Therefore, individuals often enter the group setting with different perspectives and

interpretations of the issues involved. Through interaction and discussion, members are confronted

with the conicting views of their colleagues and must seek to reconcile dissimilar assumptions under-

lying the issues. Cognitive consensus refers to similarity among group members regarding how key

issues are dened and conceptualized (Mohammed and Ringseis, in press).


Although most of the decision making research has focused on how groups negotiate to reach consen-

sus on decisions, much less is known about how group members negotiate to reach cognitive consensus

on the interpretations of issues (e.g., Brehmer, 1976; Bettenhausen, 1991). Nevertheless, an integral

part of the group effort is dedicated to resolving differences in how members conceptualize problems,



and without active consideration given to conicts caused by cognitive factors, there is not likely to be

any conict resolution (Brehmer, 1976). Researchers representing different elds of study have exam-

ined various aspects of cognitive diversity in groups. For example, the psychology literature has

focused on frames of reference in terms of gain/loss perspectives and disagreements in cue weighting

(e.g., Brehmer, 1976; Tindale et al., 1993). More diverse in its treatment of cognitive diversity, the

management literature has examined variables such as attributions concerning organizational success

and failure (Kilduff et al., 2000), belief structures (Walsh et al., 1988), and issue certainty, controll-

ability, and scope (Fiol, 1994).

Group members may attempt to resolve these types of diversity and develop collective representations

of decision issues. Cognitive consensus refers to similarity among group members regarding how key

issues are dened and conceptualized. Groups that have more cognitive consensus are likely to attend

to, interpret, and communicate about issues more similarly than individuals who have less cognitive

consensus (Mohammed and Ringseis, in press). Although decision making tasks have traditionally

focused on outcomes, cognitive consensus primarily attends to characteristics of individual judgment pro-

cessing. Whereas decision preferences reveal what members want out of the decision process, cognitive

representations help to explain the reasons underlying the preferences (e.g., Tindale et al., 1993).

Depending on the situation, scholars contend that both cognitive diversity and cognitive consensus

can contribute to effective performance in decision making groups (e.g., Kilduff et al., 2000; Walsh

et al., 1988). For example, multiple member perspectives have been shown to contribute to creative

solutions, but may also cause problems due to miscommunication and disorganization (e.g., Jackson,

1992). Therefore, cognitive diversity can assist a group in operating as a unied structure, but becomes

a liability when the uniqueness of individual contributions is lost. Because extreme diversity and con-

sensus in collective representations are generally viewed as dysfunctional for many situations, a deli-

cate balance of both agreement and disagreement is required (e.g., Fiol, 1994). However, the optimal

level of consensus and dissensus in framing perspectives that will contribute to effective group process

dynamics and decision outcomes will depend upon a number of factors, including the specic envir-

onment in which a group operates, the level of interdependence among members, the nature of the task,

and where the group is in the decision making process.

Empirical research

Scattered throughout various literatures, several authors have explored the notion of cognitive consen-

sus utilizing different terminology and focusing on diverse aspects of the phenomenon. For example,

Walsh et al. (1988) found that realized coverage (the range of belief structures voiced during a discus-

sion) and realized consensus (level of similarity among team members' individual belief structures)

were systematically related to product and rm performance by student groups participating in a

business simulation. On the other hand, Tindale et al. (1993) showed that variations in gain/loss

perspectives among group members inuenced decision processes.

Operationalizing cognitive consensus as shared assumptions underlying decision issues,

Mohammed and Ringseis (in press) investigated the antecedents, processes, and consequences of

cognitive consensus. Utilizing student groups participating in a multi-issue decision making exercise,

results revealed that unanimity decision rule groups achieved more cognitive consensus than majority

rule groups. In addition, groups whose members inquired concerning the reasons underlying others'

decision preferences, accepted others' viewpoints as legitimate, and incorporated others' perspectives

into their own interpretations of the issues arrived at a greater degree of cognitive consensus than

groups whose members engaged in less of these behaviors. With regard to consequences, cognitive

consensus positively inuenced expectations regarding decision implementation and satisfaction.



Kilduff et al. (2000) found that the most successful teams of managers participating in a business

simulation started out with diversity among members regarding attributions of organizational success

and failure, but developed more consensus over time. Therefore, cognitive diversity at the beginning of

a decision making task, integrated with cognitive consensus by the end of the task, is one way for teams

to foster both equivocality and mutual understanding. Also emphasizing the need for both a `diver-

gence and convergence of meanings' (p. 404), Fiol (1994) concluded that agreement around a broad

frame of interpretations provided the common meaning needed to move toward action in new venture

teams, regardless of differing views on issue content.

Overview

The concept of cognitive consensus provides a valuable means for understanding how decision makers

collectively make sense of ill-structured issues in a group setting and is conceptually appealing

because it integrates group, cognitive, negotiation, and decision making literatures. However, the

extant body of research devoted to the construct is fragmented and scattered across various literatures.

Recent work by Mohammed (1997 Paper presented at the Academy of Management Meeting,

Boston, MA) has attempted to consolidate and integrate what has been learned about cognitive con-

sensus into a foundation upon which future work can be based, but there are still many unanswered

questions that must be addressed both conceptually and empirically. Particularly salient is the need for

insight into processes that contribute to the development of cognitive consensus and how groups can

achieve an optimal mix of diversity and consensus in group-level interpretations.

Cognitive consensus and team mental models

Whereas both teamwork and taskwork are implicated in the notion of a team mental model, cognitive

consensus narrows the focus to collective representations of key issues faced by the group. In addition,

cognitive consensus references evaluate belief structures, while information sharing, transactive

memory, group learning, and team mental models reference knowledge structures. In contrast to

knowledge structures, which concern descriptive states of nature that one knows or thinks to be true,

belief structures concern desired states of nature that one prefers or expects (Mohammed et al., 2000).

Rather than focusing on raw information content, cognitive consensus deals with the interpretation

of the information, how it is viewed by the group, and the opinions that are held about it. Therefore,

it is likely that knowledge structures will occur more readily than the development of cognitive

consensus. Furthermore, `correctness' or accuracy is an important variable in the study of knowledge

structures, but would be problematic with cognitive consensus because of its subjective and evaluative

nature.

Although the existing work on team mental models has referenced shared knowledge structures,

the cognitive consensus literature can assist in expanding the construct to include belief structures

as well. We would argue that the team mental model construct should allow for the notion of shared

interpretations and viewpoints of strategic issues. Not only is it important that team members share an

adequate knowledge of taskwork and teamwork, but that they also have a common conception of the

assumptions underlying issues of signicance. In addition to including belief structures, team mental

model research could also benet from the work on cognitive consensus by incorporating key political

realities inherent in many group contexts. Despite the fact that political mechanisms are often over-

looked in team research (Walsh et al., 1988), recognizing the presence of divergent viewpoints, as well

as conicting motives to both compete and cooperate, could enrich team mental model research.



Development of an Integrative Framework

In examining the constructs of team mental models, information sharing, transactive memory, group

learning, and cognitive consensus, this paper reects the variety of perspectives that emerge when mul-

tidisciplinary scholars examine the broad concept of team knowledge. Collectively, these literatures

provide considerable insights into shared cognition processes at the group-level. However, the diver-

sity also provides a considerable challenge in that there are many ways to view the multifaceted nature

of team knowledge.

Table 1 summarizes the team knowledge domains that have been reviewed in the current paper

according to their academic roots, denitions, methodologies, antecedents, and criterion variables.

Unfortunately, these literatures have progressed in parallel with little cross-fertilization, and research-

ers do not commonly cite each other's work. However, because all of these areas are in the formative

stages of research development, there is much to be gained from integration across disciplinary bound-

aries. Therefore, the current paper seeks to enhance and expand understanding of team mental models

by reviewing and assimilating these separate research streams on team knowledge.

Integrating across all of the literature reviewed, one of the primary distinctions separating the

research on team mental models from the research on information sharing, transactive memory, group

learning, and cognitive consensus is that the former references teams, while the latter references

groups. Specically, team mental models are generally discussed in the context of command and con-

trol or military teams, where the task is highly structured, roles are clearly differentiated, and coordi-

nated patterns of interdependence are specied (e.g., Cannon-Bowers et al., 1993; Mathieu et al.,

2000). On the other hand, the other team knowledge domains have primarily been implicated in

groups, which are characterized by ambiguous task contexts, unspecied roles, and the absence of

explicit task interaction demands (Salas et al., 1992). As the previous review demonstrates, much

of the work on groups has been conducted in laboratories with undergraduate students doing tasks

(e.g., murder mystery; making origami birds) that are far removed from the complex, multidimen-

sional context typical of `real world' military, medical, or process control teams. Therefore, caution

should be exercised in applying theories and empirical ndings based on groups to teams, and the inte-

grative sections of the paper have taken these important differences into account when drawing con-

nections to team mental model research.

With that qualier in mind, however, we do feel that group literature, with its longer history and

greater volume, has much to offer team research. Indeed, although the empirical research has been

primarily limited to groups in laboratory settings, there is nothing inherent in the phenomenon of infor-

mation sharing, transactive memory, group learning, and cognitive consensus that prevents their

applicability to teams. For example, recent work by Lewis (2000 Paper presented at the Academy

of Management Toronto, Canada) has shown that transactive memory systems predict the performance

of consulting teams. Also breaking away from the traditional lab-based research, group learning beha-

viors have been measured in organizational teams (Edmondson, 1999; Cannon and Edmondson, 2000.

Paper presented at the Academy of Management Conference, Toronto, Canada). Hopefully, these

applied research trends will continue.

Several themes have emerged from the review of the various team knowledge literatures. First,

based on how the various constructs have been conceptualized and operationalized, it appears that

the notion of a team mental model is the broader term that encompasses the specic dimensions of

information sharing, transactive memory, group learning, and cognitive consensus. Second, these var-

ious team knowledge domains feature different content domains, such as taskwork, teamwork, and

belief systems. Third, the concepts reect varying degrees of emphasis with regard to the denition

of `shared' as overlapping/held in common versus complementary/distributed. Figure 1 portrays



Table 1. Summary of team knowledge literature

Team Academic Roots Construct/ Methodology Antecedents Criterion variablesKnowledge Phenomenon Investigated ExaminedDomain Description Empirically Empirically

Team mental Human factors Team members' shared, organized Flight simulations Team training Team processes (e.g.,model Industrial/ understanding of knowledge about (e.g., TANDEM) Planning communication,

organizational key elements in the team's relevant Similarity ratings, coordinationpsychology environment (e.g., teamwork, pathnder, UCINET, Team performance

Cognitive psychology taskwork) concept mappingInformation Social psychology Collective sampling dynamics Intellective hidden Temporal Free recall

sharing Communication promote the discussion of shared prole task (e.g., environmentinformation at the expense of murder mystery) Leadership, status Number of times unshared/pooling unshared information Laboratory-based Decision making shared information

training mentioned/repeatedAccountability Consensus decisionTask environment

Transactive Social psychology Cognitively interdependent system Radio assemblage Communication Memory differentiation/memory Cognitive psychology for encoding, storing and retrieving task in groups media recall

information that combines the Memory recall task in Imposed/natural Task coordinationknowledge possessed by individual stranger/couple dyads transactive memory Task credibilitymembers with a shared awareness Laboratory-based system Complexity, accuracy,of who knows what Individual/group agreement about

training distribution of expertiseTurnover

Collective Organizational Processes and outcomes through Laboratory and eld Task complexity Group performancelearning behavior which groups acquire and combine research designs and structure Group learning curve

new knowledge through experience Quantitative and Team safetyqualitative Leadershipmethodologies Turnover

Cognitive Decision making Similarity in the way that group Laboratory and eld Decision rule Group decisionsconsensus Organizational members conceptualize and research designs Constituencies Anticipated ease of

behavior dene key issues Quantitative and Cooperativeness decision implementationNegotiation qualitative Satisfaction with decisionsPsychology methodologies Simulation performance

10

2S

.M

OH

AM

ME

DA

ND

B.

C.

DU

MV

ILL

E

Co

py

righ

t#

20

01

Joh

nW

iley&

So

ns,

Ltd

.J.

Org

an

iz.B

eha

v.2

2,

89

1

06

(20

01

)

these themes by positioning the team mental model construct within a larger team knowledge

framework.

As previously discussed, the information sharing and transactive memory literatures have clearly

been task-oriented, focusing on the discussion of information and knowledge concerning members'

domains of expertise. Within the context of heterogeneous teams in which distinct roles require infor-

mation unique to particular individuals, it is argued that knowledge concerning the task should primar-

ily be distributed among team members. Although there is some taskwork knowledge that all team

members will need to hold in common in every team, too much overlap will create redundancies

and inefciencies that result in sub-optimal use of team resources. Therefore, the majority of taskwork

information should be unique to individual roles or be held in common among specied dyads or triads

within the team.

In contrast to the information sharing and transactive memory literatures, the group learning area has

emphasized the overlapping denition of sharing. It is argued that knowledge reecting how the

team will function together and communication processes should primarily be held in common by

all team members. Given the distributed nature of taskwork knowledge recommended in hetero-

geneous teams, overlapping teamwork knowledge would be necessary to provide adequate coordi-

nation for the team to function smoothly as a collective entity. Uneven distribution of teamwork

knowledge would undermine the development of accurate expectations of team behavior and effective

team coordination. Therefore, group learning efforts involving all members of the team should be

oriented toward dimensions of teamwork in heterogeneous teams.

Whereas teamwork and taskwork reect knowledge structures, varying interpretations and perspec-

tives regarding key team issues reect belief structures. Although team mental models have tradition-

ally measured knowledge structures, it is argued that the construct should allow for the notion of

evaluative belief structures, and the work on cognitive consensus can assist in this regard. A delicate

balance between both overlapping and complementary sharing perspectives is needed with respect to

belief structures. With completely divergent belief structures, group member interactions will likely

involve a high degree of miscommunication and misunderstanding. However, because multiple per-

spectives are one of the advantages of a group context, overlap in team member interpretations to

the point of groupthink becomes a liability. Therefore, group members must simultaneously agree

and disagree in order to maintain both unity and diversity in equilibrium. Members of the group

may even be in agreement on the need to disagree, respect divergent perspectives, and permit conict.

TEAM MENTAL MODEL CONSTRUCT SPACE

Figure 1. Team mental models within a team knowledge framework



Furthermore, within a single group, there may be coalitions of shared beliefs, with all individuals shar-

ing some beliefs, but only a sub-set of members sharing other beliefs. The optimal level of consensus

and dissensus in framing perspectives that will contribute to effective outcomes will depend upon a

number of moderating variables such as where the group is in the decision making process.

With regard to recommendations for future research, the integrative framework proposed in Figure 1

highlights the need for team mental model researchers to clearly articulate the particular content

domain being considered and the sharing perspective hypothesized. Given the multifaceted nature

of the construct, team mental models should not be referenced in the abstract without specifying

whether the focus is on teamwork/taskwork/representations of issues, and whether a distributed/com-

mon notion of sharing is being considered. In addition, the degree to which a team's mental model

spans across both knowledge and belief structure domains should be incorporated as a relevant

criterion variable in research. Whereas some teams may have a well developed task mental model,

a poorly developed teamwork mental model, and wide divergence on belief structures, other teams

may experience just the opposite.

Based on the literature reviewed in this paper, we would expect that a team mental model with pri-

marily distributed taskwork knowledge, primarily overlapping teamwork knowledge, and a balance of

diversity and consensus in belief structures would result in higher collective performance for hetero-

geneous teams where distinct team roles require information unique to particular individuals. How-

ever, for other types of teams and tasks, a different conguration may be superior. The optimal

degree of sharing across content domains will depend upon factors such as the specic environment

in which a team operates, the level of interdependence among members, the nature of the task, and

where the group is in terms of development.

Acknowledgement

The contributions of the authors were supported by a grant from the Naval Air Warfare Center Training

Systems Division in Orlando, FL.

Author biographies

Brad Dumville is a graduate student in industrial/organizational psychology at Penn State University.He received his BA in psychology from Tulane University. His research interests include organiza-

tional culture, White privilege/culture, team mental models, and methodological issues.

Susan Mohammed is an Assistant Professor of industrial/organizational psychology at Penn StateUniversity. She received her PhD in I/O psychology at Ohio State University. Her research focuses

on teams and decision making, with a specic emphasis on shared cognition, team mental models,

and team composition.

References

Adelman L, Zirk DA, Lehner PE, Moffett RJ, Hall R. 1986. Distributed tactical decision making: conceptualframework and empirical results. IEEE Transactions on Systems, Man, and Cybernetics 16: 794805.



Argote L, Gruenfeld D, Naquin C. 1999. Group learning in organizations. In Groups at Work: Advances in Theoryand Research, Turner ME (ed.). Lawrence Erlbaum: Hillsdale, NJ.

Argote L, Insko CA, Yovetich N, and Romero AA. 1995. Group learning curves: The effects of turnover and taskcomplexity on group performance. Journal of Applied Social Psychology 25: 512529.

Bettenhausen KL. 1991. Five years of groups research: what we have learned and what needs to be addressed.Journal of Management 17: 345381.

Blickensderfer E, Cannon-Bowers JA, Salas E. 1997. Theoretical bases for team self-correction: fostering sharedmental models. Advances in Interdisciplinary Studies of Work Teams 4: 249279.

Bowers CA, Jentsch F, Salas E, Braun CC. 1998. Analyzing communication sequences for team training needsassessment. Human Factors 40: 672679.

Brehmer B. 1976. Social judgment theory and the analysis of interpersonal conict. Psychological Bulletin 83:9851003.

Cannon-Bowers JA, Salas E, Converse S. 1993. Shared mental models in expert team decision making.In Individual and Group Decision Making, Castellan NJ (ed.). Lawrence Erlbaum Associates: Hillsdale, NJ;221246.

Cooke NJ, Salas E, Cannon-Bowers JA, Stout RJ. 2000. Measuring team knowledge. Human Factors 42: 151173.Daft RL, Weick KE. 1984. Toward a model of organizations as interpretive systems. Academy of Management

Review 9: 284295.Duncan PC, Rouse WB, Johnston JH, Cannon-Bowers JA, Salas E, Burns JJ. 1996. Training teams working in complex

systems: a mental model-based approach. Human / Technology Interaction in Complex Systems 8: 173231.Edmondson AC. 1999. Psychological safety and learning behavior in work teams. Administrative Science

Quarterly 44: 350383.Fiol CM. 1994. Consensus, diversity, and learning in organizations. Organization Science 5: 403420.Gigone D, Hastie R. 1993. The common knowledge effect: information sharing and group judgment. Journal of

Personality and Social Psychology 65: 959974.Hollingshead AB. 1996a. Information suppression and status persistence in group decision making: the effects of

communication media. Human Communication Research 23: 193219.Hollingshead AB. 1996b. The rank-order effect in group decision making. Organizational Behavior and Human

Decision Processes 68: 181193.Hollingshead AB. 1998a. Retrieval processes in transactive memory systems. Journal of Personality and Social

Psychology 74: 659671.Hollingshead AB. 1998b. Communication, learning, and retrieval in transactive memory systems. Journal of

Experimental Social Psychology 34: 423442.Jackson SE. 1992. Consequences of group composition for the interpersonal dynamics of strategic issue

processing. Advances in Strategic Management 8: 345382.Kilduff M, Angelmar R, Mehra A. 2000. Top management team diversity and rm performance: examining the

role of cognitions. Organization Science 11: 2134.Kleinman DL, Serfaty D. 1989. Team performance assessment decision making. In Proceedings of the Symposium

on Interactive Networked Simulation for Training, Gibson et al. (eds). University of Central Florida: Orlando,FL; 2227.

Klimoski R, Mohammed S. 1994. Team mental model: construct or metaphor? Journal of Management 20:403437.

Kraiger K, Wenzel LH. 1997. Conceptual development and empirical evaluation of measures of shared mentalmodels as indicators of team effectiveness. In Team Performance Assessment and Measurement, Brannick MT,Salas E, Prince C (eds). Lawrence Erlbaum Associates: Mahwah, NJ; 6384.

Larson JR Jr, Foster-Fishman PG, Keys CB. 1994. Discussion of shared and unshared information in decision-making groups. Journal of Personality and Social Psychology, 67: 446461.

Larson JR Jr, Christensen C, Abbott AS, Franz TM. 1996. Diagnosing groups: charting the ow of information inmedical decision-making teams. Journal of Personality and Social Psychology 71: 315330.

Larson JR Jr, Foster-Fishman PG, Franz TM. 1998. Leadership style and the discussion of shared and unsharedinformation in decision making groups. Personality and Social Psychology Bulletin 24: 482495.

Liang DW, Moreland R, Argote L. 1995. Group versus individual training and group performance: the mediatingrole of transactive memory. Personality and Social Psychology Bulletin 21: 384393.

Mathieu JE, Heffner TS, Goodwin GF, Salas E, Cannon-Bowers. 2000. The inuence of shared mental models onteam process and performance. Journal of Applied Psychology 85: 273283.

Miller D. 1996. A preliminary typology of organizational learning: synthesizing the literature. Journal ofManagement 22: 485505.



Mohammed S, Ringseis E. in press. From cognitive diversity to cognitive consensus in group decision making:The role of inputs, processes, and outcomes. Organizational Behavior and Human Decision Processes.

Mohammed S, Klimoski R, Rentsch J. 2000. The measurement of team mental models: we have no sharedschema. Organizational Research Methods 3: 123165.

Moreland RL. 2000. Transactive memory: learning who knows what in work groups and organizations. In SharedCognition in Organizations: The Management of Knowledge, Thompson L, Messick D, Levine J (eds).Lawrence Erlbaum: Hillsdale, NJ; 331.

Rentsch JR, Hall RJ. 1994. Members of great teams think alike: a model of team effectiveness and schemasimilarity among team members. Advances in Interdisciplinary Studies of Work Teams 1: 223261.

Rouse WB, Cannon-Bowers JA, Salas E. 1992. The role of mental models in team performance in complexsystems. IEEE Transactions on Systems, Man, and Cybernetics 22: 12961308.

Rulke DL, Rau D. 1997. Examining the encoding process of transactive memory in group training. Academy ofManagement Proceedings, Boston, MA; 349353.

Salas E, Dickinson TL, Converse SA, Tannenbaum SI. 1992. Toward an understanding of team performance andtraining. In Teams: Their Training and Performance, Swezey RW, Salas E (eds). Ablex Publishing Corporation:Norwood, NJ; 329.

Stasser G. 1991. Pooling of unshared information during group discussion. In Group Process and Productivity,Worchel S, Wood W (eds). Sage Publications, Inc: Newbury Park, CA.

Stasser G, Stewart DD. 1992. The discovery of hidden proles by decision making groups: solving a problemversus making a judgment. Journal of Personality and Social Psychology 63: 426434.

Stasser G, Titus W. 1985. Pooling of unshared information in group decision making: Biased informationsampling during discussion. Journal of Personality and Social Psychology 48: 14671478.

Stasser G, Titus W. 1987. Effects of information load and percentage of shared information on the dissemination ofunshared information during group discussion. Journal of Personality and Social Psychology 53: 8193.

Stasser G, Taylor LA, Hanna C. 1989. Information sampling in structured and unstructured discussions of three-and six-person groups. Journal of Personality and Social Psychology 57: 6778.

Stasser G, Stewart DD, Wittenbaum GM. 1995. Expert roles and information exchange during discussion: theimportance of knowing who knows what. Journal of Experimental Social Psychology 31: 122.

Stewart DD, Stasser G. 1998. The sampling of critical, unshared information in decision-making groups: The roleof an informed minority. European Journal of Social Psychology 28: 95113.

Stout RJ, Cannon-Bowers JA, Salas E, Milanovich DM. 1999. Planning, shared mental models, and coordinatedperformance: An empirical link is established. Human Factors 41: 6171.

Tindale RS, Sheffely S, Scott LA. 1993. Framing and group decision-making: do cognitive changes parallelpreference changes? Organizational Behavior and Human Decision Processes 55: 470485.

Walsh JP, Henderson CM, Deighton J. 1988. Negotiated belief structures and decision performance: an empiricalinvestigation. Organizational Behavior and Human Decision Processes 42: 194216.

Walsh JP, Ungson GR. 1991. Organizational memory. Academy of Management Journal 16: 5791.Wegner DM. 1987. Transactive memory: a contemporary analysis of the group mind. In Theories of Group

Behavior, Mullen IB, Goethals GR (eds). Springer-Verlag: New York; 185208.Wegner DM. 1995. A computer network model of human transactive memory. Social Cognition: 13: 319339.Wegner DM, Erber R, Raymond P. 1991. Transactive memory in close relationships. Journal of Personality and

Social Psychology 61: 923929.Wittenbaum GM, and Stasser G. 1996. Management of information in small groups. In What's Social about

Cognition, Nye JL, Brower AM (eds). Sage Publications: Thousand Oaks, CA.



86_ftp team mental models in a team knowledge

Documents